Technical Report for Soccernet 2023 -- Dense Video Captioning
This work addresses the problem of automated video analysis for soccer, providing a domain-specific solution for action captioning and timestamping.
The authors tackled dense video captioning for soccer videos by generating captions for each action and locating their timestamps, achieving results through a method combining Blip for captioning with multi-size sliding windows, temporal proposal generation, and proposal classification for timestamp localization.
In the task of dense video captioning of Soccernet dataset, we propose to generate a video caption of each soccer action and locate the timestamp of the caption. Firstly, we apply Blip as our video caption framework to generate video captions. Then we locate the timestamp by using (1) multi-size sliding windows (2) temporal proposal generation and (3) proposal classification.